PC and Smartphone Replacement Cycles

Shipments Follow Refresh Cycles

PC and smartphone unit sales rise and fall with replacement timing. A cluster of devices bought in one boom year comes due for refresh together; when that wave passes, shipments can fall even if the installed base is still large. The modelling job is to turn that history into a unit curve. Trend growth sets the slow path and a cycle index multiplies around it. The installed base then rolls forward on an average life.

Two replacement-life facts can anchor the lives used in that build. Counterpoint Research put the smartphone replacement cycle at 43.0 months in 2024. IDC described US commercial PCs as replaced every 3.5 to 4 years (2023 Technology Spotlight).

Worldwide PC Shipments

Gartner January releases, calendar years 2016-2025 (millions of units):

The 2021 peak of 339.8M fell to 241.9M in 2023, a 28.8% decline. By 2025 shipments had recovered to 270.2M, still below the 2021 peak.

Worldwide Smartphone Shipments

IDC releases on Business Wire cover calendar years 2018-2022 on a consistent primary-source basis (millions of units):

From 2019 to 2022 the series fell 12.1% (1,371.0M to 1,205.5M). The 2021 rebound did not hold into 2022.

From Cycle Facts to a Unit Curve

A practical device build has three moving pieces:

1. Trend units. A slow growth rate for the category (population, computing intensity), separate from the cycle.
2. Cycle index. A yearly multiplier around 1.00 that rises into a refresh peak and falls into a trough. PC history shows a 28.8% peak-to-trough move from 2021 to 2023; smartphones show a 12.1% decline from 2019 to 2022. Those amplitudes size how violent the index can be.
3. Installed base. Opening base, plus new units, minus retirements at 1 / average life. Smartphones at about 43 months imply a life near three and a half years; US commercial PCs at 3.5-4 years sit in the same neighbourhood. Recurring services then ride that base (Hardware Product vs Services Gross Margin).

Device brands such as Apple and PC sellers such as Dell Technologies feel the unit swing first in product revenue. Margins on cyclical hardware without a large services cushion need a mid-cycle read (Mid-Cycle Margins for Hardware Companies). Build plans sized at a peak also show up later as purchase commitments (Purchase Commitments vs Inventory in Hardware).

What the History Is For

The shipment tables are not a forecast. They show that a single year of units is a poor stand-in for normal demand when the category just came off a 28.8% PC decline or a 12.1% smartphone decline. A model that ignores replacement timing will treat a trough as permanent or a peak as a new floor.

Use the lives (43.0 months for smartphones; 3.5-4 years for US commercial PCs) to retire the installed base and the shipment amplitudes to stress the cycle index. Keep category data (Gartner for PCs, IDC for smartphones) labelled by source and calendar year.